Chapter 4. Where and what is (t,d)? 125
3.1 (t,d) and CSPs
The phonetic evidence surveyed in this paper has demonstrated that where direct
comparison is possible word-final (t,d) consonants exhibit the same patterns of
variability as other word-final stops, including variable pre-consonantal release
characteristics and a range of degrees of lenition, crucially including full (auditory/
acoustic) deletion. They also show parallel patterns of interaction with adjacent
consonants resulting from Connected Speech Processes such as assimilation and
cophonation. Such parallels have also been observed by Browman & Goldstein
(e.g. 1990) using articulatory data from X-ray pellet-tracking: cluster-final /t/ and
/d/ in perfect memory and nabbed most, when auditorily deleted, manifest a similar
residual alveolar tongue gesture to word-final /n/ assimilating to following /p/ in
seven plus. Moreover, even where direct comparisons across different consonants
are not possible, there appears to be a plausible explanation in terms of CSPs for
the whole range of variability observed in (t,d) clusters, including the behaviour
of the first consonant of the cluster.
Furthermore, if (t,d) is a manifestation of general word-final CSPs, we would
expect evidence of the cooccurrence of other CSPs in the surrounding speech.
Thus (75) above shows voicing assimilation of the schwa of you to the preceding
coalesced, devoiced [ʃ] and following /k/, the type of cophonation Nolan focuses
on (Nolan 1996: 223). Cooccurence of CSPs is illustrated more starkly in (76),
where the whole sequence except the last word, leave shows decreased phonetic
explicitness: the comments in §2.6 focussed on the coalescence at the end of want
but in fact the whole sequence is highly reduced, and he didn’t want me to being
pronounced [əndji ̃mw̰ɒw̰̃ iːtʰə]. [d] and [j] are clearly articulated sequentially,
but [n] and [d] are heterosyllabic, suggesting that the [d] is part of a coalesced
pronunciation of he di-; the first [i] is nasalised in anticipatory assimilation to
the following /n/, which assimilates in place to the /w/ of want; that /w/ is itself
creaky-voiced, suggesting it bears a reflex of the final /t/ of didn’t. In (81)^27 there
is no acoustic or auditory evidence of any alveolar closure in the whole sequence
/nt͡ʃt/, close alveolar approximation not appearing until the following consonant
/ð/. Note that nasality is also absent:
(81) so they pinched the [piːʲʃðə] typewriter
The (artificial) borderline between coalescence and cooccurrence breaks down
at this point, but as noted at the outset, these categorisations are a descriptive
convenience rather than a theoretical taxonomy. More importantly, the fact that
2 7. The (t,d) cluster in (81) would again be excluded from a variationist analysis because of the
following ‘neutralisation’ context.